General election: How well did the polls do in predicting the result?

Pollsters’ report cards favourable, but precise number of seats proves harder to pin down

As Britain’s opposition parties begin the sad task of picking up the pieces after a devastating 2019 general election result, just how accurate was the polling we saw in the run-up to Thursday’s ballot?

Despite much talk of tactical voting providing the path to thwarting Boris Johnson, the Conservatives stormed to victory by securing 44 per cent of the national vote, giving the prime minister a majority government and a clear mandate to force through his “get Brexit done” agenda whatever the consequences.

Jeremy Corbyn’s Labour Party could not compete and were left with just a 32 per cent share as the map of the British Isles turned blue, their cause arguably greatly harmed by Liberal Democrat candidates splitting the vote in key constituencies.

That certainly appeared to be the case in areas like Kensington, in west London, where Tory Felicity Buchan beat Labour’s Emma Dent Coad by just 150 votes as Conservative defector Sam Gyimah came in third for the Lib Dems, eating up crucial potential support for the incumbent Dent Coad in the process.

Nationally, the Liberals took 12 per cent of the vote, with the mythic yellow surge failing to materialise and party leader Jo Swinson losing her seat and her job in the process.

Nicola Sturgeon’s Scottish National Party (SNP) meanwhile took four per cent, the Greens three per cent and Nigel Farage’s Brexit Party just two per cent in the latest electoral humiliation for the pint-supping poster boy of Britain’s exit from the European Union.

According to the BBC poll tracker, which brings together a range of top polls to give an overall predicted outcome, the verdict on this year’s pollsters is highly positive.

Taken together, the polls had just a one per cent margin of error and only slightly overestimated support for Mr Farage’s outfit, correctly foreseeing Tory strength in areas that voted for Leave in 2016.

The BBC singles out Opinium and Ipsos Mori for particular praise when it came to accurately forecasting the final result.

What proved harder to predict was precisely how many seats each party would win.

The Tories ended up taking 365, Labour 203, the SNP 48, the Lib Dems 11, the Democratic Unionist Party eight and others picking up the remaining 15.

YouGov introduced a technique called MRP (multi-level regression and post-stratification) in 2017 that correctly forecast a hung parliament on its debut outing but was broadly felt to have been less reliable this time, underestimating the sheer extent of Conservative support.

“Although there are doubtless lessons to be learnt, this election has not been the big embarrassment for the polling industry that both the last two polls proved to be,” said Sir John Curtice, professor of politics at Stathclyde University, in The Times.

“It is the Brexit division that in turn gave rise to the unfamiliar pattern whereby the Tories advanced more strongly in seats with relatively large numbers of working-class voters than in more middle-class constituencies, and made more progress in much of the north of England and the midlands than in the south.

“It was as though someone had picked up the electoral map of Britain and turned it upside down.”